HumAInFlow : a no-code platform for modelling and simulating Human-AI workflows

The increasing integration of Generative AI (GenAI) agents into socio-technical systems, calls for platforms that can model, simulate, and analyse workflows involving both automated and human tasks. Existing agentic AI frameworks largely focus on automation and remain tightly bound to specific SDKs, often lacking structured support for human-in-the-loop modelling and simulation. To address these limitations, we introduce HumAInFlow, a no-code platform for modelling and simulating socio-technical workflows that explicitly integrates human roles as first-class nodes and supports their simulation via large language models (LLMs). The platform is SDK-agnostic, supports both local and remote LLM execution, and integrates the Model Context Protocol (MCP), ensuring interoperability and extensibility. Comparative analysis shows that HumAInFlow advances the state of the art by combining privacy-preserving deployment, execution monitoring, reproducibility, and explicit support for human–AI collaboration.